A new murine chemokine was identified in a search for glucocorticoid-attenuated response genes induced in the lung during endotoxemia. The first 73 residues of the predicted mature peptide are 71% identical and 93% similar to human CXCL11/IFN-inducible T cell α chemoattractant (I-TAC) (alias β-R1, H174, IFN-inducible protein 9 (IP-9), and SCYB9B). The murine chemokine has six additional residues at the carboxyl terminus not present in human I-TAC. Identification of this cDNA as murine CXCL11/I-TAC is supported by phylogenetic analysis and by radiation hybrid mapping of murine I-TAC (gene symbol Scyb11) to mouse chromosome 5 close to the genes for monokine induced by IFN-γ (MIG) and IP10. Murine I-TAC mRNA is induced in RAW 264.7 macrophages by IFN-γ or LPS and is weakly induced by IFN-αβ. IFN-γ induction of murine I-TAC is markedly enhanced by costimulation with LPS or IL-1β in RAW cells and by TNF-α in both RAW cells and Swiss 3T3 fibroblasts. Murine I-TAC is induced in multiple tissues during endoxemia, with strongest expression in lung, heart, small intestine, and kidney, a pattern of tissue expression different from those of MIG and IP10. Peak expression of I-TAC message is delayed compared with IP10, both in lung after i.v. LPS and in RAW 264.7 cells treated with LPS or with IFN-γ. Pretreatment with dexamethasone strongly attenuates both IFN-γ-induced I-TAC expression in RAW cells and endotoxemia-induced I-TAC expression in lung and small intestine. The structural and regulatory similarities of murine and human I-TAC suggest that mouse models will be useful for investigating the role of this chemokine in human biology and disease.
The chemokines are a large superfamily of cytokines involved in the regulation of leukocyte trafficking and other functions (1, 2, 3, 4, 5, 6, 7). To date, about 50 human chemokines have been described. The biological activities of chemokines are mediated via interaction with seven-transmembrane domain G-protein-coupled receptors. Most chemokines are small, secreted proteins of molecular mass 7–10 kDa that bind specifically to heparin. The vast majority belong to two main subfamilies defined by the positions of two pairs of conserved cysteines. Disulfide bonds link the first with the third cysteine and link the second with the fourth. In the CXC subfamily, the first two cysteines are separated by a single variable amino acid (X), whereas in the CC subfamily, the first two cysteines are adjacent. The CXC chemokines are further subdivided into two main groups based on structural and functional correlations. Seven human CXC chemokines (IL-8; neutrophil-activating peptide-2; growth-related oncogene (GRO)3/melanoma growth stimulatory activity-α, -β, and -γ; epithelial cell-derived neutrophil-activating peptide-78; and granulocyte chemotactic protein-2) are potent neutrophil chemoattractants. These chemokines all contain a glutamic acid-leucine-arginine (ELR) motif immediately preceding the first conserved cysteine and are referred to as ELR+ CXC chemokines. Seven other CXC chemokines (platelet factor 4, B lymphocyte chemoattractant/B cell-attracting chemokine 1, stromal cell-derived factor 1, BRAK, IFN-inducible protein 10 (IP10), monokine induced by γ-IFN (MIG), and IFN-inducible T cell α chemoattractant (I-TAC; human CXCL11)) lack the ELR motif and are inactive toward neutrophils (7, 8).
Three human chemokines (I-TAC, MIG, and IP10) constitute a structurally and functionally related subset of genes within the non-ELR CXC chemokine subgroup. Although human I-TAC (alias β-R1, H174, IP-9, or SCYB9B) was identified only recently (9, 10, 11, 12, 13, 14), human and murine IP10 and MIG were among the first chemokines to be discovered (15, 16, 17, 18). Both IP10 and MIG are chemotactic for activated but not resting T cells (19), and both are ligands for the chemokine receptor CXC chemokine receptor 3 (CXCR3) (20). CXCR3 is mainly expressed on activated Th1 cells and NK cells and therefore may play a critical role in selectively recruiting these cells to inflammatory sites (21, 22, 23, 24, 25, 26). Subsequent down-regulation of CXCR3 may also be important in allowing these cells to recirculate to lymph nodes (27). CXCR3 is expressed on infiltrating lymphocytes in active multiple sclerosis lesions (28, 29) and on the T lymphocytes within human atherosclerotic lesions (30). Although not present on normal B cells, CXCR3 is also expressed on malignant B cells in chronic lymphocytic leukemia and some other lymphoproliferative disorders (31). The expression characteristics of CXCR3 suggest that this receptor and its ligands could be important targets for therapeutic blockade in a variety of diseases.
The human I-TAC cDNA was identified by large-scale sequencing of clones from stimulated human astrocytes, and its predicted product was shown to be a potent chemoattractant for IL-2-stimulated T cells, acting via CXCR3 (11). A partial cDNA (β-R1), not initially recognized as a chemokine, had previously been isolated as an IFN-γ-induced gene in astrocytes (9). The H174 cDNA was obtained by a genetic selection for secreted proteins (10), whereas the SCYB9B cDNA was discovered serendipitously as part of a chimeric sequence (12, 13). The native protein (IP-9) was isolated in a search for novel ligands of CXCR3 secreted by IFN-γ-stimulated keratinocytes (14). Human I-TAC (gene symbol SCYB11, formerly SCYB9B) maps to chromosome 4q21.2, in close proximity to MIG (SCYB9) and IP10 (SCYB10) (12, 32). The murine orthologues of MIG (MuMig) and IP10 (crg-2, C7) are ligands for murine CXCR3 (16, 17, 33, 34, 35), and map to mouse chromosome 5 (36). The murine orthologue of human I-TAC has not been described previously.
The anti-inflammatory effects of glucocorticoid hormones are mediated, in part, by their ability to attenuate the induction of message expression of genes encoding inflammatory mediators, including numerous cytokines such as TNF-α, IL-1, and IL-8 and enzymes such as the inducible forms of prostaglandin synthase and nitric oxide synthase (37, 38). We refer to this class of genes as glucocorticoid-attenuated response genes (GARGs) (39, 40). We hypothesized that there are many GARGs not yet identified, and in a previous study we cloned a new murine CXC chemokine, LPS-induced CXC chemokine (gene symbol Scyb5), by screening a cDNA library prepared from Swiss 3T3 fibroblasts for LPS-induced, dexamethasone-attenuated messages (39). We recently extended GARG screening to an in vivo model designed to identify endotoxemia-induced, dexamethasone-attenuated genes expressed in the lungs of adrenalectomized mice. In the course of this project, we identified a cDNA fragment representing an unknown gene. In this article we describe cDNAs containing the complete coding sequence of this gene. Sequence comparisons, phylogenetic analysis, gene mapping, and regulatory properties support the identification of this gene as the murine orthologue of human I-TAC.
Materials and Methods
Cycloheximide, dexamethasone, murine IFN-αβ, and sterile, tissue-culture certified LPS prepared by phenol extraction and gel filtration from Escherichia coli serotype O111:B4 were obtained from Sigma (St. Louis, MO). Preservative- and pyrogen-free saline (Abbott Laboratories, North Chicago, IL) was used for dilution of LPS and for control injections. Dexamethasone sodium phosphate (4 mg/ml of dexamethasone phosphate equivalent) for injection was obtained from Elkins-Sinn (Cherry Hill, NJ). Recombinant murine IFN-γ was obtained from Sigma and from R&D Systems (Minneapolis, MN). Recombinant murine IL-1β and TNF-α were obtained from R&D Systems.
Male Swiss Webster mice, purchased from Charles River Laboratories (Cambridge, MA) or from Simonson Laboratories (Gilroy, CA), were studied at 8–12 wk of age. Groups of mice used in individual experiments were obtained from a single supplier. Adrenalectomized mice from Charles River Laboratories were used in library preparation and screening. Adrenalectomized mice were given normal saline instead of water to drink and were studied 2–4 wk after adrenalectomy. Other experiments reported in this paper utilized nonoperated mice, except as noted. Mice were housed in specific pathogen-free conditions in the UCLA Center for the Health Sciences Vivarium. Experiments were conducted in accordance with a protocol approved by the UCLA Animal Research Committee.
LPS (50 μg in 200 μl sterile saline or saline alone) was administered i.v. via the tail vein. In experiments involving dexamethasone, two 400-μg doses (or 100 μl saline) were administered s.c. 16–20 h before and 5 min before LPS injection. At specified intervals after LPS injection, the mice were anesthetized with halothane and were then killed by cervical dislocation. The pulmonary artery was flushed with 3 ml cold PBS before removal of the lung lobes, which were individually dissected free of the surrounding soft tissues.
Cell culture media were from Life Technologies (Rockville, MD), and FBS was from Omega Scientific (Tarzana, CA). RAW 264.7 cells were cultured in 5% CO2/air in high-glucose DMEM with 10% FBS and antibiotics. Murine Swiss 3T3 cells were cultured at 37°C in 5% CO2/air in low-glucose DMEM supplemented with 10% FBS and antibiotics. Near-confluent cultures were switched to medium containing 0.5% FBS for 18–24 h before induction with IFN-γ or other agents.
RNA preparation and Northern analysis
Whole organs or tissue segments of ≤250 mg dissected from the mice were immediately homogenized in 3 ml of RNA extraction buffer and were frozen at −80°C for subsequent RNA isolation by the acid phenol method (41). RNA from cultured cells was prepared using Trizol (Life Technologies). Electrophoresis of total cellular RNA (10 μg/lane), transfer, and hybridization conditions were as described (39). Autoradiographic exposures were made at −80°C with one intensifying screen using XAR film (Kodak, Rochester, NY). Expression of I-TAC, MIG, and IP10 was determined either by sequential stripping and probing of the same membrane or by probing replicate filters made from the same RNA. Each filter used was then stripped and reprobed with a cDNA probe for the murine ribosomal S2 protein to assess variations in loading. Quantitation was performed using phosphor imaging (Molecular Dynamics, Sunnyvale, CA). Corrections for variations in loading were made using the signal intensity of S2.
Library synthesis and screening
A cDNA library was prepared using lung tissue harvested from adrenalectomized mice 1, 2, or 4 h after tail vein injection of 50 μg LPS. Total cellular RNA extracted from the lungs of individual mice was checked for absence of degradation by Northern analysis and then pooled (8–10 mice per time point). Poly(A)-RNA was isolated using the polyATtract kit (Promega, Madison, WI). cDNA was prepared using the Superscript II cDNA Synthesis kit (Life Technologies). First-strand synthesis was primed with oligo-dT. After second-strand synthesis, ligation of Not-Sal-EcoRI adapters, and size selection, the cDNA was ligated into the EcoRI site of λ ZapII and packaged with Gigapack III Gold (Stratagene, La Jolla, CA). The primary library, containing 1 × 106 recombinants, was amplified once. The library was screened by hybridization with probe from the 068D fragment (see Results), and insert sizes of plaque-purified candidate phage were determined by PCR as described (40), except that amplification was performed using vector primers ATTAACCCTCACTAAAGGGA and TAATACGACTCACTATAGGG with an annealing temperature of 58°C. Selected phages were then converted to plasmid form (42) and sequenced.
Sequence and phylogenetic analysis
Editing of sequences, assembly into contigs, and open reading frame analysis were performed using AssemblyLign 1.09b and MacVector 6.5 (Oxford Molecular, Oxford, U.K.). Multiple sequence alignment and phylogenetic analysis were performed using Clustal W (43, 44). The unrooted tree diagram was generated from the Clustal W output with Treeview 1.5.3 (45). Analysis of signal cleavage sites was performed using Signal-P (46, 47).
Radiation hybrid mapping
DNA samples from the 100 cell lines of the T31 radiation hybrid panel, which carries fragments of the mouse genome on a hamster background (48), were used as templates for PCR amplification with murine I-TAC primers CCTGGGAACGTCTGACTGTG and GAAGGTAGCGTGGAGTGTGC. Preliminary experiments showed robust amplification of the expected 195-bp product (nt 370–564 of Fig. 1) from the cloned I-TAC cDNA. No products were obtained when the murine MIG and IP10 cDNAs were used as templates. Reactions were performed with 112.5 ng of hybrid clone DNA, 0.4 μM of each primer, 250 nM of each dNTP, and 0.625 U of TaKaRa Ex Taq (Takara Shuzo, Otsu, Japan) and the supplied buffer in a total volume of 25 μl. After an initial denaturation at 94°C for 3 min, 30 cycles of denaturation at 94°C for 30 s, annealing at 62°C for 1 min, and extension at 72°C for 1.5 min were performed before a final extension at 72°C for 7 min. The PCR products were analyzed by electrophoresis in 1.2% agarose gels, stained with ethidium bromide, and scored for the presence or absence of the 195-bp product. Two independent reactions were performed with the I-TAC primers for the entire T31 panel. Additional reactions were performed to clarify discordant results for I-TAC and to verify the location of markers determined to be near the I-TAC (Scyb11) locus genetically. All data were submitted to The Jackson Laboratory Mouse Radiation Hybrid Database (http://www.jax.org/resources/documents/ cmdata/).
Plasmids and probes
Probes were prepared by random-primed synthesis using α-32P-deoxycytidine 5′-triphosphate from NEN (Boston, MA). Templates were gel-purified fragments excised from cloned cDNA. The murine I-TAC probe was a 1.2-kb EcoRI fragment from clone 068D14. This fragment was used, rather than the full-length insert, to avoid the repetitive sequences present in the 3′ end of the cDNA (see Fig. 1). The murine MIG cDNA was a gift from Dr. Joshua M. Farber (National Institutes of Health, Bethesda, MD) (33). The IP10 (crg-2/GARG-10), macrophage-inflammatory protein-2 (MIP-2), and S2 clones we used were described previously (39, 49).
Cloning of the murine I-TAC cDNA
The murine I-TAC cDNA was identified in a search for glucocorticoid-attenuated response genes induced in the lung during endotoxemia. A two-stage strategy was used in this search. First, we prepared a subtracted library enriched in endotoxemia-induced genes and then screened this library by differential hybridization for genes whose induction in endotoxemia was attenuated by pretreatment with dexamethasone (J.B.S., unpublished observations). Briefly, cDNA subtraction was performed using the suppression subtractive hybridization method (50, 51). The “tester” cDNA population was prepared from mRNA isolated from lungs of adrenalectomized mice 1, 2, and 4 h after i.v. injection of LPS. The “driver” cDNA population was prepared from lung mRNA from adrenalectomized mice treated with dexamethasone. The tester cDNA was digested with RsaI, ligated to suppression subtractive hybridization adapters 1 and 2, hybridized with RsaI-digested driver, and reamplified by suppression PCR as described (50, 51). The result of this procedure was a population of cDNA fragments highly enriched for endotoxemia-induced genes (data not shown). A λ ZapII library prepared from this subtracted cDNA population was then screened by differential hybridization to select glucocorticoid-attenuated messages (40). A candidate cDNA fragment designated 068D contained a sequence not found in GenBank.
To further characterize the 068D partial sequence, we constructed a full-length (i.e., not RsaI-digested), unsubtracted cDNA library using poly(A)+ RNA from lungs of adrenalectomized, LPS-treated mice as described in Materials and Methods, and screened this library for clones hybridizing with the 068D insert. Two independent clones containing the identical complete coding sequence were isolated. Fig. 1 shows the sequence of the longer clone. The shorter cDNA started at nt 20 and terminated with a 21-nt poly(A) tail following nt 1032 of the sequence shown. This site follows overlapping rare polyadenylation signals AATATA and TATAAA at nt 1006–15. The polyadenylation signal AATAAA is present at nt 914–19, but we did not isolate any clones with a 3′ end corresponding to this site. A recent analysis indicates that the “canonical” AATAAA signal is actually utilized in only about 25% of mouse messages (52). With 200–400 nt poly(A) tails, messages terminating after either the 914–19 or the 1006–15 potential polyadenylation signals would be compatible with the predominant 1.2- to 1.4-kb band we observed on Northern blots. A second band with about 5% of the signal intensity of the major band was consistently seen at ∼4.5 kb (not shown).
Sequence similarities and phylogenetic analysis
The open reading frame of the murine I-TAC cDNA encodes a predicted 100-amino acid proprotein with an N-terminal signal sequence and a mature peptide containing four cysteines in the positions characteristic of the CXC chemokine family (Fig. 1). Analysis of predicted signal peptide cleavage sites using Signal-P (46) showed nearly equal probabilities of cleavage between Ala19 and Gln20 (Y score, 0.578) and between Gly21 and Phe22 (Y score, 0.573). For the human I-TAC sequence in contrast, Signal-P unambiguously predicts cleavage between positions 21 and 22 (Y score, 0.817; the second highest Y score is 0.321 for cleavage between 20 and 21), in agreement with the N-terminal sequence of the protein isolated from human keratinocyte supernatants by Tensen et al. (14). If the processing of the murine I-TAC proprotein is similar to that of human I-TAC and the other closely related chemokines, cleavage would occur after Gly21 (Fig. 2). The resulting 79-aa residue mature murine I-TAC would have a molecular mass of 9112 Da. If cleavage occurred after Ala19 instead, the mature protein would have 81 residues and a mass of 9297 Da. With either cleavage site the predicted protein is highly basic (isoelectric point, 10.8).
The murine and human I-TAC proproteins are highly similar (Fig. 2). The 73 aligned residues of the predicted mature peptides are 71% identical and 93% similar, whereas the signal sequences are 57% identical and 63% similar (Fig. 2). Murine I-TAC has six additional residues at the carboxyl terminus not present in the human protein. Next to human I-TAC, the other chemokines most closely related to murine I-TAC are human and murine IP10 (32–36% identical residues in the aligned portions) and human and murine MIG (25–29% identical). Thus, the murine I-TAC protein has much greater similarity to human I-TAC than to any other chemokine, murine or human. The IP10 and MIG sequences also form closely related murine/human pairs, whereas the overall similarity among the six I-TAC, IP10, and MIG sequences is much less than the similarity within each pair (Fig. 2). A phylogenetic analysis of all known murine and human CXC chemokines clearly identifies the protein encoded by the cDNA we cloned as the murine orthologue of human I-TAC and shows that I-TAC, MIG, and IP10 form a distinct subgroup of chemokines in which all three genes have highly conserved murine and human orthologues (Fig. 3).
Murine I-TAC maps to mouse chromosome 5, close to IP10 and MIG
The chromosomal location of murine I-TAC was determined using the T31 mouse/hamster radiation hybrid panel (48) as described in Materials and Methods. Based on the mapping and sequence data, murine I-TAC was assigned the gene symbol Scyb11 by the International Committee on Standardized Genetic Nomenclature for Mice (The Jackson Laboratory, Bar Harbor, ME). The radiation hybrid data place Scyb11 on mouse chromosome 5 between markers D5 Mit20 and D5 Mit369 (Fig. 4). The logarithm of odds scores were 5.6 between D5 Mit20 and Scyb11, and 9.9 between D5 Mit369 and Scyb11.
For two of the D5 Mit markers tested, the physical and genetic mapping data are inconsistent. By radiation hybrid analysis, D5 Mit369 and D5 Mit274 are separated by only 20 cRay and are located between D5 Mit20 and D5 Mit89, which map to 52 and 53 cM from the centromere (Fig. 4). However, by backcross analysis, D5 Mit369 and D5 Mit274 are widely separated and in the reverse order, mapping to 68 cM and 45 cM, respectively. In contrast, the physical and genetic mapping orders are entirely consistent for D5 Mit312, D5 Mit20, D5 Mit89, D5 Mit10 (Fig. 4), and other nearby markers (not shown). Based on these markers, the physical map position for murine Scyb11 corresponds to a locus at 52–53 cM on the genetic map, as shown in Fig. 4. Murine IP10 and MIG have previously been mapped to ∼53 cM by backcross analysis, whereas murine KC and MIP-2 map to ∼51 cM (36). This portion of mouse chromosome 5 is syntenic to human chromosome 4q12–21, which contains the human CXC chemokine cluster. Human SCYB11 (I-TAC), IP10, and MIG are located within ∼30 kb at 4q21.2, separated by ∼2 Mb from the CXC chemokine minicluster at 4q12 (32). Our radiation hybrid mapping data for murine I-TAC (Scyb11), combined with the linkage data for KC, MIP-2, IP10, and MIG (36), indicate that murine I-TAC, like its human orthologue, is located in close proximity to the IP10 and MIG genes.
Murine I-TAC is induced in multiple tissues during endotoxemia, with a time course and pattern of tissue expression different from those of MIG and IP10
Our candidate insert for what proved to be I-TAC was cloned as an LPS-induced, glucocorticoid-attenuated cDNA from lung. To confirm the inducibility of I-TAC and to assess its expression in other tissues, we evaluated message expression of I-TAC in tissues from LPS-treated and control mice, in comparison with MIG and IP10. I-TAC message was not detected in control tissues but was induced after LPS injection in all tissues examined (Fig. 5). Induced I-TAC expression was greatest in lung, heart, small intestine, and kidney. In contrast, MIG was most strongly induced in heart and liver, whereas IP10 was most strongly induced in lung, spleen, and kidney. Taking account of the differences in exposure times for the autoradiograms in Fig. 5, it is apparent that IP10 message is considerably more abundant than I-TAC and MIG in most tissues of the LPS-treated mice at this time point. (The probes used were similar in length (1.1–1.3 kb) and were labeled to similar specific activities.)
To determine whether the time course of I-TAC induction in endotoxemia differs from MIG and IP10, we evaluated message expression in the lung 1–8 h after LPS injection (Fig. 6). I-TAC message was first detectable 2 h after LPS and was still increasing at 8 h. MIG induction was even more delayed. In contrast to I-TAC and MIG, message expression of IP10 was substantial at 2 h and peaked at 4 h. Nevertheless, the initial increase in IP10 expression was delayed compared with the neutrophil-chemoattractant CXC chemokine MIP-2, whose message expression was maximal at 1 h. The differing time courses and patterns of tissue expression of I-TAC, MIG, and IP10 indicate that these chemokines are individually regulated.
Dexamethasone attenuates the induction of I-TAC, MIG, and IP10 during endotoxemia
To verify that I-TAC is a glucocorticoid-attenuated response gene and to quantitate the magnitude of the attenuation, we examined the effect of dexamethasone on endotoxemia-induced message expression in the lung and small intestine. Dexamethasone (or saline) was given 16–20 h before and 5 min before i.v. injection of LPS, and tissues were harvested 4 h later. Three mice were studied for each condition. Dexamethasone reduced LPS-induced I-TAC message expression by 85% in lung and by 93% in small intestine (not shown). In the same mice, dexamethasone reduced LPS-induced MIG and IP10 expression by 80% or more in both lung and small intestine. Dexamethasone also reduced LPS-induced expression of all three chemokines by 80% or more in lung and small intestine of adrenalectomized mice (not shown). Thus, dexamethasone strongly attenuates induction of I-TAC, MIG, and IP10 in this endotoxemia model in both nonoperated and adrenalectomized mice.
It is interesting to note that clones for IP10 were abundantly represented among the candidate phages we isolated in our screening for glucocorticoid-attenuated genes (J.B.S., unpublished observations). However, we did not isolate any MIG clones, most likely because MIG expression is so low (Fig. 6) at the 2 h time point utilized in the screening.
Murine I-TAC expression is induced by IFN-γ and by LPS in RAW 264.7 macrophages and is attenuated by dexamethasone
In RAW 264.7 macrophages, murine I-TAC was strongly induced by as little as 10 IU/ml IFN-γ but was only weakly induced by 1000 IU/ml IFN-αβ (not shown). In agreement with previous work (16, 17, 33), MIG was induced by IFN-γ but not by IFN-αβ, whereas IP10 was strongly induced by either IFN-αβ or IFN-γ (not shown). Although I-TAC, MIG, and IP10 were all inducible by IFN-γ in RAW 264.7 cells, the kinetics of their responses to IFN-γ were distinctly different (Fig. 7,A). IP10 message was rapidly induced and peaked at 4 h, as described previously (53). I-TAC message expression did not begin to rise substantially until after 2 h and peaked 6–8 h after IFN-γ (Fig. 7,A and data not shown). IFN-γ-induced MIG induction was even more delayed, with peak message expression at 12 h. For each gene, message expression at 24 h (not shown) remained close to the 16-h level. In RAW 264.7 cells, LPS induced the message expression of I-TAC and IP10 but not of MIG (Figs. 7,B and 8A). For both I-TAC and IP10, the time courses of message expression in response to LPS and to IFN-γ were similar (Fig. 7, A and B). Interestingly, the relative kinetics of the responses to IFN-γ in RAW cells (prompt for IP10, intermediate for I-TAC, and delayed for MIG) were identical with what we observed in the lung during endotoxemia (Fig. 6). The in vivo response, of course, might be partially due to the direct effects of LPS and partially due to the effects of endogenous mediators released in endotoxemia, including TNF-α, IL-1β, and IFN-γ.
In RAW 264.7 cells, dexamethasone markedly attenuated the IFN-γ-induced message expression of I-TAC, MIG, and IP10 (Fig. 7 C). Expression of each chemokine message was reduced by 80% or more at 4 h, a level of attenuation similar to what we observed at 4 h in vivo, and by more than 90% at 8 h.
Inhibition of protein synthesis with cycloheximide did not block IFN-γ induction of I-TAC message expression in RAW 264.7 cells (Fig. 7 D). MIG and IP10 also were induced in the presence of cycloheximide, as previously observed (16, 33). This indicates that the induction of these genes by IFN-γ is not dependent on synthesis of new transcription factors or other intermediates. Thus, all three of these chemokines are immediate-early or primary response genes (54), despite the delayed peak expression of I-TAC and MIG compared with that of IP10. However, secondary factors could participate in the continuing rise of I-TAC and MIG message expression after 4 h.
IFN-γ induction of murine I-TAC is enhanced by costimulation with LPS, IL-1β, or TNF-α
In RAW 264.7 cells, message induction of both I-TAC and MIG in response to IFN-γ together with LPS was increased more than 8-fold compared with induction by IFN-γ or LPS alone (Fig. 8,A). This synergistic effect was particularly striking for MIG, which was not detectably induced by LPS alone. Strong synergistic effects were seen for all three genes in response to the combination of IFN-γ with IL-1β, with TNF-α, or with IL-1β plus TNF-α (Fig. 8,A). In Swiss 3T3 cells, induction of I-TAC, MIG, and IP10 by IFN-γ was much weaker than in RAW cells, but the combination of IFN-γ with TNF-α or with IL-1β plus TNF-α strongly induced all three messages (Fig. 8,B). For all three genes, message induction by TNF-α in combination with IFN-γ was enhanced more than 40-fold compared with induction by TNF-α or IFN-γ alone. Interestingly, the weak IFN-γ-induced message expression of I-TAC, MIG, and IP10 in Swiss 3T3 cells was dramatically enhanced by cycloheximide: the IFN-γ-induced message levels of these genes were increased 35-fold or more in the presence of cycloheximide but were not increased by cycloheximide in the absence of IFN-γ (not shown). The synergistic effects of the combination of IFN-γ with LPS, IL-1β, or TNF-α that we observed for murine I-TAC, MIG, and IP10 (Fig. 8) have also been reported for the human orthologues of these genes in several cell types (11, 13, 30, 55, 56, 57). These synergies might be particularly significant in the early stages of Th1-dependent processes, when few IFN-γ-secreting T cells have arrived at a local site of inflammation and local IFN-γ concentrations are low.
The recent description of human I-TAC as a chemoattractant for activated T cells has stimulated great interest in the potential role of this chemokine in T-cell-mediated processes. Human I-TAC, like MIG and IP10, is induced in a variety of cell types by IFN-γ, and each of these chemokines is a ligand for CXCR3 (11, 14, 20). Because CXCR3 is predominantly expressed on activated Th1 cells (21, 22, 23, 24, 25, 26), which are characterized by their ability to secrete IFN-γ and IL-2, all three chemokines may contribute to Th1-type immune responses by recruiting CXCR3-expressing Th1 cells to a local site of inflammation. The IFN-γ secreted by these Th1 cells would stimulate increased local production of IP10, MIG, and I-TAC, which in turn would recruit additional Th1 cells. This self-amplifying local paracrine loop could have an important role in the establishment and maintenance of Th1-type responses (20, 25).
The identification of murine I-TAC allows us to compare the structures of all three of the human CXCR3 ligands (I-TAC, MIG, and IP10) with their murine counterparts. The predicted mature murine I-TAC protein has much greater similarity to human I-TAC (71% identical, 93% similar residues) than to the most closely related murine chemokines, MIG and IP10 (29–36% identity). The sequence similarities between murine and human MIG, and between murine and human IP10, are equally close (Fig. 2). The phylogenetic analysis (Fig. 3) and the mapping of murine I-TAC to mouse chromosome 5 near the murine MIG and IP10 genes (Fig. 4) support the idea that I-TAC, MIG, and IP10 arose by gene duplication from a common precursor early in mammalian evolution (32, 58). In addition, the preservation of both high sequence similarity and distinct regulatory features (discussed below) within each murine-human pair of homologues suggests that the I-TAC, IP10, and MIG genes may have developed individually conserved functions before the divergence of the rodent and primate lineages. In considering this suggestion, it is useful to note that a clear one-to-one correspondence between human and murine genes does not exist for all chemokines. The neutrophil-chemoattractant, ELR+ subgroup of CXC chemokines provides several examples (Fig. 3), including 1) the lack of a murine orthologue of human IL-8, 2) the likely origin of human epithelial cell-derived neutrophil-activating peptide-78 and granulocyte chemotactic protein-2 from an evolutionarily recent gene duplication of an ancestral gene orthologous to murine LPS-induced CXC chemokine (LIX) (59, 60), and 3) the lack of identifiable orthologues for any of the individual human GRO genes, though as a group the GRO genes are closely related to murine KC and MIP-2 (58). These examples suggest that there has been ample evolutionary time for correspondences between murine and human chemokines that originated from common ancestral genes to have become obscured. Thus, the preservation of high similarity between human and murine I-TAC, between human and murine MIG, and between human and murine IP10 (Figs. 2 and 3) supports the idea that each of these genes may have an evolutionarily conserved function.
The existence of multiple ligands for a single receptor is a common theme in the chemokine superfamily (3, 7). Differential regulation in distinct cells types or tissues is one of several mechanisms by which chemokines that act on a common receptor could have functionally distinct properties in vivo (7, 25, 61). Our studies of the endotoxemia model provide two lines of evidence for differential regulation of I-TAC, MIG, and IP10 in vivo. First, the patterns of tissue expression of these genes are different. I-TAC message, undetectable in control tissues, is most abundantly expressed during endotoxemia in lung, heart, small intestine, and kidney, a pattern quite different from those of MIG and IP10 (Fig. 5). In contrast to I-TAC, endotoxemia-induced MIG expression is greatest in liver. This is consistent with a previous study showing that liver expression of MIG is prominent during the acute phase of various viral and protozoal infections and after IFN-γ injection (62). The pattern of IP10 expression in that study varied for different stimuli (62). In endotoxemia, we found that IP10 expression was prominently induced in lung and kidney and, in direct contrast to I-TAC, was high in spleen but low in small intestine (Fig. 5). Except for the observation that IP10 is induced in liver and kidney after LPS injection (63), tissue expression of IP10 and MIG has not previously been evaluated in endotoxemia, so our study provides new information for these chemokines as well as for I-TAC. Second, we found that the time courses of I-TAC, MIG, and IP10 induction are different. Lung expression of I-TAC and MIG in response to LPS injection is delayed compared with that of IP10. Message expression of IP10 rises abruptly at 2 h and peaks 4 h after LPS, whereas I-TAC expression and MIG expression continue to increase between 4 and 8 h. However, for I-TAC the rate of rise between 4 and 8 h was modest compared with that of MIG, suggesting that the expression of MIG peaks later than that of I-TAC, which is consistent with the in vitro kinetics we found for these genes (Fig. 7).
Studies of human tissues have also provided evidence for differential expression of I-TAC, MIG, and IP10. Skin biopsies of allergic contact reactions and mycosis fungoides show consistent topographic expression patterns by in situ hybridization, with I-TAC expressed mainly in the epidermis, MIG mainly in the dermis, and IP10 in both (14). In human carotid atherosclerotic plaques, smooth muscle cells express IP10 and MIG but not I-TAC, whereas endothelial cells and macrophages express all three chemokines (30). Such differences in localization of expression are consistent with the idea that I-TAC, MIG, and IP10 may have nonredundant roles in vivo (19, 30). Experimental models showing a specific requirement for I-TAC or IP10 have not been reported. However, a recent study suggests that T cell infiltration and rejection of class II MHC-disparate allografts in mice may be specifically dependent on intra-allograft production of MIG (64).
To examine further the differential regulation of I-TAC, MIG, and IP10, we compared their responses to induction by IFN-γ, IFN-αβ, and LPS in RAW 264.7 macrophage cells. Murine I-TAC, MIG, and IP10 are all strongly induced by IFN-γ in these cells (Figs. 7 and 8). However, the three genes have differing responses to IFN-αβ and to LPS. Murine I-TAC is weakly induced by IFN-αβ (not shown) and modestly induced by LPS (Fig. 8,A) relative to IFN-γ. Similarly, human I-TAC is weakly induced by IFN-α or IFN-β in THP-1 myelomonocytic cells and strongly induced by IFN-γ (13). However, induction of human I-TAC by LPS as a single stimulus was not observed in THP-1 cells (13). In contrast to I-TAC, we found that murine IP10 is strongly induced in RAW cells by IFN-αβ and by LPS (as well as by IFN-γ), whereas MIG is induced only by IFN-γ (Figs. 7 and 8), which is in agreement with previous studies of IP10 and MIG in both murine and human monocyte/macrophage cell lines (15, 18, 19, 53). Thus, each of the three related chemokines has a different pattern of responsiveness to stimulation with IFN-γ, IFN-αβ, and LPS, and with the exception of the I-TAC response to LPS, the pattern for each chemokine is similar in both murine and human monocyte/macrophage cell lines.
Although induction of human I-TAC by LPS as a single stimulus was not observed in THP-1 cells (13), costimulation with LPS enhances IFN-γ induction of human I-TAC in these cells (13). In fact, the combination of IFN-γ with LPS, IL-1β, or TNF-α produces synergistic increases in human I-TAC, MIG, and IP10 induction in a wide variety of cell types (11, 13, 30, 55, 56, 57). We found that costimulation with LPS, IL-1β, or TNF-α also markedly augments IFN-γ induction of murine I-TAC in RAW 264.7 macrophages, Swiss 3T3 fibroblasts, or both cell types (Fig. 8). Murine IP10 and MIG were also synergistically induced, as previously reported (65, 66, 67). For murine I-TAC, we observed the greatest synergy in Swiss 3T3 cells, in which the combination of TNF-α with IFN-γ induced more than 50-fold greater expression than either agent alone (Fig. 8 B and data not shown).
Glucocorticoids are widely used in the treatment of inflammatory diseases. Among other mechanisms, inhibition of the expression of proinflammatory cytokines is thought to be an important component of the anti-inflammatory effects of glucocorticoids (37, 38, 68, 69). For human I-TAC, glucocorticoid effects on induction so far have been investigated in only one study, which found that dexamethasone had little or no effect on induction of human I-TAC, MIG, or IP10 in bronchial epithelial cells stimulated with IFN-γ or with IFN-γ in combination with TNF-α or IL-1β (55). In contrast, we found that dexamethasone strongly attenuates the induction of murine I-TAC, as well as MIG and IP10, in both lung and small intestine during endotoxemia and in IFN-γ-stimulated RAW macrophages (Fig. 7 C), as previously observed for IP10 in LPS-stimulated RAW cells (70). These observations suggest that glucocorticoid attenuation of I-TAC, MIG, and IP10 expression may contribute to the anti-inflammatory effects of glucocorticoids in Th1-dependent processes.
The I-TAC, MIG, and IP10 genes have been highly conserved between mouse and human, and message regulation of each gene is similar in both species. The CXCR3 gene has also been highly conserved between mouse and human and is preferentially expressed in activated Th1 cells in both species (34, 35). Murine CXCR3 is activated by human I-TAC, murine MIG, and murine IP10 and, as in the human system (11, 14), the hierarchy for cross-desensitization of murine CXCR3 is human I-TAC > MIG > IP10 (34, 35). Although murine 6Ckine induces a Ca2+ flux via murine CXCR3 (34) whereas human 6Ckine does not induce a Ca2+ flux via human CXCR3 (71), the significance of this difference is uncertain because murine 6Ckine fails to induce chemotaxis in murine CXCR3 transfectants (35). Thus, the available data suggest that the CXCR3 receptor/ligand system functions similarly in mice and humans and that mouse models (interpreted with appropriate caution (71)) should be valuable for studying the role of I-TAC and the other CXCR3 ligands in human immunity and disease.
We thank Profs. Harvey R. Herschman and Oto Martínez-Maza for helpful discussions and comments on the manuscript and Dr. Joshua M. Farber for the murine MIG cDNA.
This work was supported by National Institutes of Health Grants HL57008 (to J.B.S.) and HL30568 (to A.J.L.).
Abbreviations used in this paper: GRO, growth-related oncogene; ELR, glutamic acid-leucine-arginine; IP10, IFN-inducible protein-10; MIG, monokine induced by IFN-γ; I-TAC, IFN-inducible T cell α chemoattractant; CXCR3, CXC chemokine receptor 3; GARG, glucocorticoid-attenuated response gene; MIP-2, macrophage-inflammatory protein-2.